Cell Cycle Analysis of Epstein-Barr Virus-Infected Cells following Treatment with Lytic Cycle-Inducing Agents

Abstract

While Epstein-Barr virus (EBV) latency-associated gene expression is associated with cell cycle progression, the relationship between the EBV lytic program and the cell cycle is less clear. Using four different EBV lytic induction systems, we address the relationship between lytic cycle activation and the cell cycle. In three of these systems, G₀ or G₁ cell growth arrest signaling is observed prior to detection of the EBV immediate-early gene product Zta. In tetradecanoyl phorbol acetate-treated P3HR1 cultures and in 5-iodo-2′-deoxyuridine-treated NPC-KT cultures, cell cycle analysis of Zta-expressing cell populations showed a significant G₁ bias during the early stages of lytic cycle progression. In contrast, treatment of the cell line Akata with anti-immunoglobulin (Ig) results in rapid induction of immediate-early gene expression, and accordingly, activation of the immediate-early gene product Zta precedes significant anti-Ig-induced cell cycle effects. Nevertheless, cell cycle analysis of the Zta-expressing population following anti-Ig treatment shows a bias for cells in G₁, indicating that anti-Ig-mediated induction of Zta occurs more efficiently in cells traversing G₁. Last, although 5-azacytidine treatment of Rael cells results in a G₁ arrest in the total cell population which precedes the induction of Zta, cell cycle analysis of the Zta-expressing population shows a significant bias for cells with an apparent G₂/M DNA content. This bias may result, in part, from activation of Zta expression following demethylation of the Zta promoter during S-phase. Together, these studies indicate that induction of Zta occurs through several distinct mechanisms, some of which may involve checkpoint signaling.

Epstein-Barr virus (EBV) is associated with a variety of cancers in humans including African Burkitt's lymphoma, Hodgkin's disease, and nasopharyngeal carcinoma (16, 21, 28), and recent studies have identified EBV in breast tumors (3, 18, 19). EBV is also a causative agent in lymphoproliferative disorders in immunocompromised individuals (16, 21).

EBV utilizes two separate classes of genes that carry out very distinct functions in its life cycle (16, 17). The latency-type gene expression patterns are associated with cell proliferation, and many of these genes function, in part, to activate cell cycle pathways leading to cell proliferation. Accordingly, some form of latency gene expression is invariably observed in EBV-associated tumors. Although the association between latency gene expression and cell cycle progression is well established, the relationship between the lytic gene expression program and the cell cycle is less well understood. Nevertheless, previous studies have provided evidence that in contrast to the latency gene expression program, the lytic replication cycle may be associated with a nonproliferating cellular milieu. In the oral epithelium, lytic replication occurs primarily in the outer, more differentiated layers (2, 26, 27). Other studies have shown that the immediate-early EBV gene product Zta can induce cell growth arrest (5–7, 22), indicating that the EBV lytic program has evolved a mechanism to shut down cellular DNA synthesis. Cell cycle regulation during activation and progression of the lytic cascade, however, has received only limited attention (5, 12). Here we report an examination of cell cycle alterations associated with induction and progression of EBV lytic replication in four distinct lytic cycle induction systems.

MATERIALS AND METHODS

Cell culture.

NPC-KT cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Life Technologies), 2 mM glutamine, 100 μg of streptomycin per ml, and 100 U of penicillin per ml in a humidified atmosphere at 37°C with 5% CO₂–95% air. The Burkitt's lymphoma cell lines Rael, Akata, and P3HR1 were cultured as indicated above except that they were propagated in RPMI medium supplemented with 10% fetal bovine serum, 2 mM glutamine, streptomycin (100 μg/ml), and penicillin (100 U/ml). Induction experiments were carried out as indicated in the figure legends.

Cell cycle analysis.

For cell analysis of whole-cell populations, cells were collected, washed once with 1× phosphate-buffered saline (PBS), suspended in 5 ml of cold (4°C) 1× PBS–0.1% glucose, fixed with 5 ml of 70% cold (−20°C) ethanol for at least 45 min at 4°C, washed with 1× PBS, and treated for 45 min at 37°C with RNase A (0.1 mg/ml) in 69 mM propidium iodide–38 mM sodium citrate. Cell cycle analysis was carried out with a FACScan fluorescence-activated cell sorter (FACS) (Becton Dickinson).

Cell cycle analysis of Zta selected cells was carried out as follows. Induction experiments were performed as indicated in the figure legends. At the indicated time points, cells were collected, washed once with 1× PBS, suspended in 0.5 ml of cold (4°C) 1× PBS–0.1% glucose, and fixed with 5 ml of 70% cold (−20°C) ethanol for at least 45 min at 4°C (and up to 1 week). Fixed cells were spun down and washed one time with 5 ml of 1× PBS. Cells were then suspended in 50 to 100 μl of 1× PBS. One milliliter of 3.7% formaldehyde was added, and the samples were mixed briefly and then incubated for 10 min with gentle rocking. The cells were diluted with 10 ml of 1× PBS, spun down, and washed two times with 1× PBS. Samples were then suspended in 1 ml of 0.2% Triton X-100 and incubated for 3 min. Then 10 ml of 1× PBS was added, and the cells were spun down and washed one time with 10 ml of 1× PBS. After taking off the supernatant, cells were spun briefly, and the remaining PBS was taken off. Cells were then suspended in 50 μl of the first antibody solution containing a 1:50 dilution of anti-Zta polyclonal antibody RR839 (a generous gift from George Miller) and 10% fetal bovine serum in 1× PBS. The samples were incubated for 1 h at room temperature with occasional mixing. Then 1 ml of 1× PBS was added, and cells were spun down. Cells were then suspended in 200 μl of second antibody solution containing a 1:50 dilution of fluorescein isothiocyanate-conjugated goat anti-rabbit immunoglobulin (Ig) antibody and 10% fetal bovine serum in 1× PBS. Samples were then incubated for 30 min at room temperature in the dark. The cells were washed once with 1 ml of 1× PBS and suspended in 300 μl of 69 mM propidium iodide–38 mM sodium citrate solution, and RNase A was added to a final concentration of 0.1 ml/ml. Cells were incubated for 30 to 60 min at 37°C and then analyzed by fluorescence-activated cell sorting.

Western blot analysis.

After a single 1× PBS wash, a fraction of cells harvested for cell cycle analysis were separated for Western blot analysis. Cells were immediately suspended in 15 pellet volumes of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer (Laemmli) (21) and boiled for 20 min to shear the DNA. Cell lysates were subjected to SDS-PAGE separation and transferred to nitrocellulose membranes. The blots were blocked for 5 to 30 min in blocking buffer (Tris-buffered saline, 0.5% Tween 20 [TBST] containing 5% lowfat powdered milk and 1% fetal bovine serum) and then incubated with the indicated primary antibody (in blocking buffer) overnight at 4°C. The blots were washed three times with 1× TBST for 15 min each. The blots were then incubated with peroxidase-conjugated secondary antibody in blocking buffer for 1 h at room temperature. Blots were washed as above and analyzed with an enhanced chemiluminescence detection system (Amersham) according to the manufacturer's recommendations, and filters were exposed to Kodak XR film. Antibodies employed for each experiment are indicated in the figure legends.

RESULTS

Cell cycle alterations in 5-iodo-2′-deoxyuridine-treated NPC-KT cells.

Initiation of the lytic cycle in NPC-KT cells by transfecting with a plasmid that constitutively expresses the immediate-early gene product Zta results in cell growth arrest in the G₀/G₁ phase of the cell cycle (5, 8). In this system, growth arrest is accompanied by induction of the tumor suppressor protein p53 as well as the cyclin-dependent kinase inhibitors p27 and p21 (5, 6; unpublished data). To extend these studies, we analyzed alterations in the cell cycle control machinery in NPC-KT cells treated with the exogenous lytic cycle-inducing agent 5-iodo-2′-deoxyuridine. As shown in Fig. 1A, addition of 5-iodo-2′-deoxyuridine to NPC-KT cell cultures causes delayed induction of Zta at 48 and 72 h. Induction of Zta expression coincides with an increase in the percentage of cells in the G₀/G₁ phase of the cell cycle (Fig. 1B), suggesting a possible relationship between these two events. Notably, there is a substantial and reproducible decrease in the G₀/G₁ population (and a corresponding increase in the G₂/M population) at 16 h, although we do not at present understand its significance. At the molecular level, several changes in cell cycle control proteins occur prior to any detectable expression of Zta (downregulation of c-Myc and induction of p53, p27, and p21). This suggests that although Zta expression correlates temporally with the onset of growth arrest in this induction system, the molecular changes that elicit growth arrest may be upstream from Zta signaling. Although we have shown previously that Zta can induce growth arrest, the data suggest the possibility that Zta expression may itself be influenced by growth arrest signaling.

FIG. 1.

Lytic cycle induced in NPC-KT cells by treatment with 75 μg of 5-iodo-2′-deoxyuridine (IDU) per ml. Cells were harvested at the indicated times, and a portion of the cells were lysed immediately in SDS-PAGE loading buffer, boiled for 20 min, and loaded onto an SDS-PAGE gel for Western blot analysis (A). The remaining fraction of cells were analyzed for cell cycle distribution by FACS analysis of propidium iodide-stained cells (B). Zta was detected using the M47 polyclonal antibody, and p53, p27, and p21 were detected using the DO-1 (Santa Cruz Biotechnology), anti-Kip1/p27 (Transduction Laboratories), and anti-Cip1/p21 (Transduction Laboratories) monoclonal antibodies, respectively.

In most EBV lytic cycle induction systems, only a fraction of the cell population typically enters the lytic cascade. Although treatment of NPC-KT cells with 5-iodo-2′-deoxyuridine leads to an increase in the percentage of cells in the G₀/G₁ phase of the cell cycle, it was important to determine whether a similar disposition occurred specifically in the cell population undergoing lytic cycle progression. To address this issue, NPC-KT cells were induced with 5-iodo-2′-deoxyuridine and the DNA content was determined specifically in the Zta-expressing cells. Importantly, previous studies in other herpesvirus systems have shown that viral replication can measurably influence the DNA content and lead to mistaken conclusion that cells are in S or G₂/M (4, 13). In an attempt to avoid this potential complication, we assayed the cell cycle distribution in Zta-positive cells at early time points following the first detection of Zta. As shown in Fig. 2, at 38 and 48 h post-5-iodo-2′-deoxyuridine treatment, both the Zta-positive population and the total population are enriched for G₀/G₁ cells. This indicates that initiation of the EBV lytic cascade occurs in a G₀/G₁-enriched cell population in this system.

FIG. 2.

Cell cycle distribution of NPC-KT cells following exposure to 75 μg of 5-iodo-2′-deoxyuridine (IDU) per ml was analyzed by FACS analysis of propidium iodide-stained cells. Zta-positive cells were selected using the polyclonal anti-Zta antibody RR839 (a generous gift from George Miller). The cell cycle experiment shown here has been repeated with similar results.

Cell cycle alterations following TPA treatment of Burkitt's lymphoma cell line P3HR1.

To further address the relationship between lytic replication and the cell cycle, we extended our study to analyze the cell cycle in B-cell lytic induction systems. As shown in Fig. 3, treatment of the Burkitt's lymphoma cell line P3HR1 with the phorbol ester tetradecanoyl phorbol acetate (TPA) results in a significant enrichment of cells in the G₀/G₁ phase of the cell cycle within 24 h of treatment (note: a portion of these G₀/G₁ cells appear to reenter the cell cycle at 48 and 72 h). Although Zta expression is induced slightly at 24 h (Fig. 3B), significant expression of Zta is not observed until the 48-h time point, indicating that growth arrest largely precedes induction of Zta in TPA-treated P3HR1 cells.

FIG. 3.

P3HR1 cells were treated with TPA (20 ng/ml) for the indicated times. Cells were harvested and analyzed for cell cycle distribution (A) and protein expression (B) as described in the legend to Fig. 1.

The TPA-induced G₀/G₁ arrest in P3HR1 cells appears to occur through a mechanism that is distinct from the growth arrest elicited by 5-iodo-2′-deoxyuridine in NPC-KT cells. At 24 h following addition of TPA, when growth arrest is maximal, no changes in p21, p27, or p53 are observed (Fig. 3B). In addition, P3HR1 cells harbor a mutant p53, further discounting a likely role for p53 in TPA-mediated growth arrest. Interestingly, although p27 levels are not altered at 24 h, there is an apparent delayed induction at 48 and 72 h. We have previously shown that Zta induces p27 in epithelial systems (5), and the temporal correlation between p27 and Zta induction suggests that Zta may play a role in the activation of p27 expression in this system.

As shown in Fig. 4, we also analyzed the cell cycle distribution in the Zta-positive population following TPA treatment of P3HR1 cells. Like the whole cell population, an enrichment of G₀/G₁ cells is also observed in the Zta-expressing population at 24 and 38 h (Fig. 4). Notably, fewer Zta-expressing cells reenter the cell cycle at 38 h compared to the percentage of the total population that reenter the cell cycle at this time point (Fig. 4). It is possible that this results from Zta-mediated growth arrest functions, which might involve p27, or from the action of some other EBV-encoded lytic gene product. Further experiments will be required, however, to adequately address this issue.

FIG. 4.

Cell cycle analysis of the whole population and Zta-positive cells following treatment of P3HR1 cells with TPA (20 ng/ml). Zta-positive cells were selected using the polyclonal anti-Zta antibody RR839. The cell cycle experiment shown here has been repeated with similar results.

Cell cycle regulation in anti-Ig-treated Akata cells.

Unlike 5-iodo-2′-deoxyuridine stimulation of NPC-KT cells and TPA stimulation of P3HR1 cells, activation of the lytic cascade through anti-Ig treatment of Akata cells has been shown previously to occur with rapid kinetics, with significant Zta induction typically observed within 4 h (25). This rapid response likely precludes the involvement of a growth arrest response as a significant factor in lytic cycle activation in this system, since such cell synchronization requires a considerably longer time frame. As shown in Fig. 5, treatment of Akata cells with anti-Ig results in induction of Zta between 4 and 8 h posttreatment and no apparent G₀/G₁ arrest is observed even through 48 h postinduction. These data indicate that, as expected, anti-Ig signaling of Zta expression likely occurs through a more direct mechanism.

FIG. 5.

Akata cells were treated with anti-Ig (1 mg/ml) for the indicated times and cells were analyzed for cell cycle (A) and protein expression (B) as described in the legend to Fig. 1.

The p53 gene has been shown previously to be deleted in Akata cells (1), and accordingly, no p53 protein was detected by Western blot analysis (data not shown). Analysis of p27 levels showed a gradual decrease between 0 and 48 h (Fig. 5B), indicating that anti-Ig treatment does not elicit p27 induction. This result also suggests either that Zta cannot activate p27 expression in this cell system or that anti-Ig treatment downregulates p27 expression and Zta cannot overcome this activity. Analysis of p21 expression, on the other hand, revealed a transient induction at 2 and 4 h post-anti-Ig treatment (Fig. 5B), and although growth arrest is not involved in signaling Zta expression in this system, it remains possible that this induction of p21 may in some way be involved in lytic cycle signaling.

We also analyzed the cell cycle distribution in the Zta-positive cell population to determine whether there might be any virus specific cell cycle effects. As shown in Fig. 6, an enrichment of cells in G₀/G₁ was observed in the Zta-positive population relative to the whole population at 4, 6, and 8 h post-anti-Ig treatment. Similarly, there is an increase in the percentage of G₂/M-phase cells in the Zta-positive population at 4, 6, 8, and 10 h. It is unlikely that this enrichment is the result of the activity of any lytic gene expression (such as Zta) on the cell cycle control machinery, since it occurs so rapidly. Instead, it is more likely that although anti-Ig treatment may activate Zta expression through a relatively direct mechanism that does not require the induction of cell growth arrest, it may nevertheless cooperate with a G₁- and G₂/M-specific factor(s) to fully activate Zta expression. Alternatively, it is also possible that there is an S-phase-specific factor that inhibits Zta expression.

FIG. 6.

Lytic cycle induced in Akata cells by incubating with anti-Ig (1 mg/ml) for the indicated times. Cell cycle distribution was analyzed in the total cell population and the Zta-positive population using the anti-Zta antibody RR839.

As shown in Fig. 6, anti-Ig treatment of Akata cells results in a significant increase in the number of cells with less than a 2n DNA content. Interestingly, this activity is strongly suppressed in the Zta-positive cell population, suggesting that a viral lytic antigen(s) functions to block the apoptotic and/or necrotic anti-Ig response. Suppression of this activity is observed at the earliest time point that Zta is detected, indicating that this viral activity must be encoded by a very early gene. Similar results have been observed by another group, who showed that an EBV lytic gene specifically inhibits apoptosis induced by lytic cycle-inducing agents (G. Inman, U. Binne, G. Parker, P. Farrell, and M. Allday, submitted for publication).

Cell cycle analysis of Rael cells following treatment with 5-azacytidine.

As shown in Fig. 7, induction of the lytic cycle in the Burkitt's lymphoma cell line Rael by treatment with 5-azacytidine results in an increase in the percentage of cells in G₁ in the whole cell population. The enrichment of G₀/G₁ cells is first observed at 12 h and is pronounced by 24 h. In this time course experiment, Zta is not detected until 24 h, suggesting that growth arrest occurs either prior to or coincident with Zta expression. Unexpectedly, however, analysis of the cell cycle distribution specifically in Zta-positive cells showed a strong enrichment for cells in G₂/M (Fig. 8). As mentioned above, studies in other herpesvirus systems have shown that viral replication can make a significant contribution to the total DNA content, leading to the mistaken impression that cells undergoing lytic replication are in the S or G₂/M phase of the cell cycle (4, 13). In an effort to avoid the possible contribution of newly replicated viral DNA in the cell cycle analysis, we carried out a more detailed time course so that we could specifically analyze the cell cycle distribution of Zta-positive cells at the earliest possible time points. As shown in Fig. 9, analysis of Zta-positive cells at the earliest time that Zta is detected showed a strong bias for cells in G₂/M. To further address whether this G₂/M enrichment is the result of possible viral replication, we carried out an induction experiment in the presence of the viral replication inhibitor foscarnet. The presence of foscarnet, however, did not influence the G₂/M bias of Zta-positive cells (data not shown). Lastly, with DNA samples prepared from the experiment shown in Fig. 9, no detectable viral DNA replication was observed even at the 28-h time point (data not shown). Together, these results indicate that the enrichment of cells with an apparent 4n DNA content specifically in the Zta-positive population is not due to viral replication.

FIG. 7.

Rael cells were incubated with 10 μM 5-azacytidine (5-AZA), and cells were analyzed for cell cycle distribution at the indicated times. Anti-Zta Western blot was performed as described in the legend to Fig. 1.

FIG. 8.

Rael cells were incubated with 10 μM 5-azacytidine (5-AzaC), and cells were analyzed for cell cycle distribution at the indicated times. Cell cycle analysis of Zta-positive cells was carried out using the anti-Zta antibody RR839. Pop., population.

FIG. 9.

Cell cycle analysis (A) of uninduced and induced whole populations (Pop.) and Zta-positive populations as described in the legend to Fig. 8. Western blot analysis was performed as described in the legend to Fig. 1.

Like the results shown above for TPA-treated P3HR1 cells, no significant change in p53 levels is observed in 5-azacytidine-treated Rael cells, and this is consistent with the mutant status of p53 in Rael cells (Fig. 9B). On the other hand, a detectable increase in p27 is observed at 12 h following the addition of 5-azacytidine, and this early induction may be the result of the treatment itself. A further increase is observed around 16 to 20 h, which may also be related to the treatment itself and/or the onset of Zta expression, which is first observed between 16 and 18 h (Fig. 9A and B). No p21 was detectable at any time, which may be due to a lack of induction or to low basal levels.

DISCUSSION

It has become clear over the last few years that unlike small DNA tumor viruses, herpesviruses have likely evolved to replicate their DNA in arrested or nonproliferating tissues (10). Moreover, studies from a number of laboratories have shown that herpes simplex virus (HSV), cytomegalovirus (CMV), and EBV all encode factors that interact with the cell cycle control machinery in a manner that activates certain cell cycle checkpoint functions (10). Although activation of these checkpoints can, in some settings, lead to an apparent G₂/M arrest, the accumulating literature suggests that these herpesviruses likely evolved to elicit predominantly a G₀ and/or G₁ growth arrest.

We have shown previously (5) that Zta activates a G₀/G₁ checkpoint (although in certain EBV-negative settings, Zta can also induce a G₂/M checkpoint response [unpublished]), supporting the notion that EBV has evolved to carry out its viral replication cycle in the G₀/G₁ phase of the cell cycle. Here we show that in several lytic cycle induction systems, initiation of the lytic cascade occurs preferentially in G₀/G₁ cell populations. Cell cycle analysis of Zta-positive cells from TPA-induced P3HR1 cells, 5-iodo-2′-deoxyuridine-treated NPC-KT cells, and anti-Ig-stimulated Akata cells showed an enrichment for cells in G₀ or G₁.

TPA treatment of P3HR1 cells and 5-iodo-2′-deoxyuridine stimulation of NPC-KT cells resulted in G₀/G₁ growth arrest signaling prior to detectable expression of Zta. Furthermore, there is a significant delay of at least 24 h before Zta expression is induced in both of these systems, which is approximately the time required for significant cell cycle synchronization. It is clear that this delay is not an intrinsic property of the virus, since Zta expression occurs with rapid kinetics in anti-Ig-treated Akata cells. Together, these data suggest the possibility that some signaling events that are specifically activated in growth-arrested cells may be involved in inducing Zta expression. As such, EBV may have evolved with this response mechanism to detect when cells are in G₀ and/or G₁ so that the virus can initiate the lytic cascade under conditions that favor efficient progression of lytic replication. Signaling of the lytic cycle by anti-Ig, on the other hand, appears to occur through a distinct mechanism that more directly activates Zta expression. Nevertheless, cell cycle analysis of the Zta selected population revealed a bias for G₁ (and G₂/M) cells, suggesting that although anti-Ig signaling may not directly involve growth arrest, it may cooperate with G₁-specific factors to activate Zta expression.

Another group has recently reported that induction of the lytic cycle in EBV-infected B lymphocytes through stimulation of CD40 similarly leads to concomitant growth arrest and induction of p27 (12), further supporting the idea that the Zta promoter may be responsive to growth arrest signaling. In other studies, analysis of Zta's promoter has identified elements that have been found to be binding sites for the terminal differentiation factor MEF2 (11, 24), and Karimi et al. (15) identified a region that is responsive to epithelial cell differentiation. Therefore, accumulating evidence suggests that the Zta promoter is specifically responsive to growth arrest and/or terminal differentiation signaling.

Immediate-early gene expression in CMV and HSV has similarly been shown to be regulated in the G₁ phase of the cell cycle. Salvant et al. (23) showed that following infection, no immediate-early gene expression is detected until the cells have entered the G₁ phase of the cell cycle. Jordan et al. (14) demonstrated that activation of HSV immediate-early gene expression by the virion component VP16 requires the presence of a G₁-specific cyclin-dependent kinase. Therefore, cell cycle-responsive immediate-early gene regulation appears to be conserved among these different herpesvirus family members.

To date, all of the herpesvirus factors that have been shown to induce cell growth arrest are virion components and/or immediate-early transcription factors (10). It appears then that herpesviruses have evolved to replicate with a specific temporal sequence of events that are intimately linked with the cell cycle. Immediate-early gene expression is specifically induced by G₀- or G₁-specific factors. Either virion growth arrest proteins and/or the newly expressed immediate-early gene products then elicit specific signaling to lock cells in the appropriate stage of G₀ or G₁. Since immediate-early gene expression also elicits parallel signaling of viral DNA replication, this coupling of early viral events to the G₀ or G₁ phase of the cell cycle may be important to help prevent immediate-early genes from activating viral DNA replication at an inappropriate point in the cell cycle.

The strong G₂/M bias of Zta-positive cells following exposure of Rael cells to 5-azacytidine is in striking contrast to the G₀/G₁ bias observed in the other lytic cycle-inducing systems discussed here. A previous study has shown that demethylation may play a role in the induction of Zta expression (9). Therefore, it is possible that induction of Zta following 5-azacytidine treatment of Rael cells occurs, at least in part, through demethylation of its promoter during S phase. In this case, Zta expression would be expected to be detected in late S phase and/or in G₂, which is consistent with the results reported here. Interestingly, however, the Zta-positive cells appear to be retained in G₂/M phase (although there is clearly some leakage, as evidenced at later times [Fig. 8 and 9]). We have previously shown that Zta induces checkpoint functions that can affect both the G₁ and G₂ checkpoints (5, 6). Furthermore, in some genetic backgrounds, Zta can preferentially induce a G2 arrest (A. Rodriguez and E. K. Flemington, unpublished). Therefore, it is possible that 5-azacytidine induces demethylation of the Zta promoter during S phase, activating Zta expression, which in turn effects a G₂ block.

Although demethylation of the Zta gene may play a role in activating its expression, it is curious that there is an apparent delay of ca. 16 h before Zta can first be detected. If demethylation of Zta's promoter fully explained the induction of Zta expression, then Zta induction would be expected to occur immediately as the fractions of the population that are in S phase progress to G₂. It is possible that hemimethylation is insufficient to induce Zta expression and that cells must cycle twice before the Zta promoter is fully demethylated and the Zta promoter becomes activated. On the other hand, it is clear that 5-azacytidine induces a G₁ arrest in a portion of the population. It is possible that these arrested cells secrete an autocrine factor that may cooperate with demethylation to activate Zta gene expression. Further experiments will be required to determine whether either of these scenarios are involved in the regulation of Zta expression in this system.